论文部分内容阅读
本文在二叉判别树的基础上,提出了一种新的多级假设检验的两级图象匹配方法。给出了表示多级假设检验的二叉判别树的定义及计算代价公式,导出了一种新的可变门限,得出了采用可变门限的多级假设检验的两级图象匹配的计算代价。计算机模拟实验表明,本文提出的方法能突破两级模板匹配计算代价的极限,降低图象匹配的计算代价,同时保证了接近于平均绝对差算法的匹配定位精度。
In this paper, based on the binary discriminant tree, a new two-level image matching method is proposed. The definition of the binary discriminant tree which represents the test of multistage hypothesis and the formula of calculating the cost are given. A new variable threshold is derived, and the calculation of two-level image matching based on multi-level hypothesis testing with variable threshold is obtained. cost. Computer simulation results show that the proposed method can overcome the limitation of the two-level template matching computational cost and reduce the computational cost of image matching, meanwhile, it can ensure the matching accuracy of the position close to average absolute difference algorithm.